The performance of an indirect ELISA for diagnosing Mycobacterium bovis infection in live badgers was evaluated by examining blood samples collected from 1982 badgers captured during statutory badger removal operations in south west England. The Validity of the test and the factors affecting the prevalence of infection are described. The sensitivity of the ELISA was 40.7 percent, its specificity was 94.3 percent, the predictive value of a positive test was 67.5% percent and the predictive value of a negative test was 84.6 percent. Its sensitivity was significantly higher in males and animals with gross lesions typical of tuberculosis. The sensitivity and positive predictive values were enhanced when the results were grouped by control operation. Variables of significance for prevalence were the county, the time of year, the age and sex of animal, and the time after the start of a control operation. The possible use of the ELISA as a screening test is discussed.